Presentation on theme: "Potentials and limitations of models of nitrogen fate in soil-crop systems at different scales Bjørn Molt Petersen Faculty of Agricultural Sciences, Institute."— Presentation transcript:
Potentials and limitations of models of nitrogen fate in soil-crop systems at different scales Bjørn Molt Petersen Faculty of Agricultural Sciences, Institute of Agroecology and Environment, University of Aarhus DENMARK
Presentation overview Incorporating knowledge from experiments into simulation models, exemplified by soil organic matter modelling Overview of the farm-scale model FASSET: structure and function Moving from topsoil measurements, to modelling the whole soil profile An essential aspect of environmental assessment in agriculture: the timescale Biogas treatment of slurry: effects on soil organic matter and nitrogen leaching Conclusions
Dynamic simulation models Technically, the kind of dynamic simulation models utilised here typically represents an integration of a large number of differential equations which cannot be solved analytically. Such dynamic models are in many cases the only way at hand to answer questions of great concern, e.g. expected levels of nitrate leaching by different policies and regulations. We can not wait for 10 – 100 field experiments with a duration of 5 – 50 years for answers, but need them here and now. The price is the associated uncertainties, but no answer still leaves us in a much worse position. As one of numerous examples, the climate projection for the future of the earth are also carried out by simulation models. Dynamic simulation models may be great tools. On the other hand, the large numbers of parameters involved and the level of complexity may in the worst case cause the models to give non-optimal results. In order to take advantage of the full potential of dynamic simulation, both the models and their results should be analysed carefully and critically, and the models should be tested and further improved to the largest possible extend.
The turnover of organic matter in the soil is central in many egro-ecological issues, as it is intimately connected with nitrate leaching and loss of greenhouse gasses. In most soil-plant models, the combined carbon (C) and nitrogen (N) turnover in the soil is in fous. For the development of CN-SIM, a large dataset was utilised for calibration and validation, consisting of 22 long-term field treatments and 43 different short term laboratory treatments. Isotope information ( 15 N, 13 C, 14 C) was utilised whenever possible. CN-SIM was coupled to a parameter optimisation software with a multiple criteria target function, a database with all experimental results and software with automated graphical representation routines. This set-up allowed the estimation of parameters on a statistical basis, and assessment of parameter confidence intervals. The model performance was validated on data not applied in the parameterisation. The soil soil organic matter model CN-SIM
Example of performance on long-term data Measurements (symbols) and simulations (lines) from the Broadbalk continuous wheat experiment. Soil carbon (C) (top) and 14 C content in % modern (bottom) from 0-23 cm depth from unfertilised (BROAD-UNF), mineral fertilised (BROAD-MIN) and farmyard manure fertilised (BROAD-FYM). From Petersen et al. (Soil Biol. Biochem. 2005). Note the very large growth in soil organic matter in the experiment with application of animal manure, and how long this growth persists – this has relevance for some coming simulation examples.
Model performance on short-term data (1) Measured replicates (symbols) and simulated (lines) CO 2 evolution and mineral nitrogen (N) from an experiment with amendment of kale leaves, oilseed rape haulm and pea haulm. From Petersen et al. (Soil Biol. Biochem. 2005). This looks fairly “nice”, so let us turn to the next slide…
Model performance on short-term data (2) Measured (symbols) and simulated (lines) CO 2 evolution and mineral N from an experiment with unamended soil, wheat straw, sheep manure and wheat stubble. From Petersen et al. (Soil Biol. Biochem. 2005). Generally, the simulations of the short-term data may be characterised as acceptable to fair, despite the large effort. So: modelling short-term C and N dynamics is much tougher than long-term dynamics!
FASSET – a whole farm model considering C and N flows and losses NH 3, N 2 O NH 3,N 2 0,N 2 Deposition Fixation Fertiliser Manure www.fasset.dk
Incorporating the soil organic matter model into FASSET FASSET, as shown, is a whole-farm model, containing a soil-plant sub- model. Below are the modelled net processes of the N turnover in one soil layer in FASSET.
Failing 1 st attempt to model the soil profile : Simulation of N-leaching When implementing the developed soil organic matter model in the FASSET whole farm model, the N leaching in initial tests turned out to be at a level above the realistic. This was quite puzzling, as this model was intended to represent state-of-the-art improvement. But put into the context of whole soil-plant-atmosphere systems with realistic contents of C and N in the soil profile they failed to simulate realistic N-leaching levels. As the simulation of N-leaching is a very important part of FASSET’s purpose, this was quite unsatisfactory!
Downward transport of organic matter? The available measurements from almost all long-term experiments are results of the balance between the input and the decay of organic matter from the topsoil. And almost all organic matter models are based on a paradigm of no interaction between soil layers. However, there is clear evidence from literature on isotope studies that there is a substantial downward transport of organic matter. So a significant proportion of the apparent decay from the topsoil may in reality just represent transport to the subsoil. With few exceptions, no models of organic matter take this fully into account.
Adjustments: A fairly crude shortcut The process of resolving mechanisms and setting parameters for yet another, further improved organic matter model that accounts for the issues of the whole soil profile will probably be quite time-consuming. In order to be able to utilise the improved organic matter within FASSET in the meantime, two coarse assumptions were made: 1. The “real” turnover rate of the “active humus” pool is 50% of that in CN- SIM, because 50% of the apparent decay is assumed to be transported to deeper soil layers 2. The “passive humus” content of the subsoil is set to 70% rather than the assumed 40.5 % in the topsoil, as judged from various subsoil 14 C measurements. The adjustment resulted in lower N mineralisation and realistic N leaching levels. So we are a bit further – in my view – but there is still work to be done. This also is a good example of the need to recognise that modelling of the soil is still a comparatively young discipline, with remaining pitfalls!
Comparing nitrogen budget calculations with dynamic modelling The following slides show an modelling example, where a crop rotation purely based on mineral fertiliser input was changed to intensive pig slurry application. This was done according to Danish regulation rules, where 100 kg nitrogen in slurry is assumed to have the same fertilisation value as 75 kg from mineral fertiliser (the substitution value also mentioned by Dr. Pedersen). The focus was on the change in nitrogen leaching rather than the absolute level of leaching. To model approaches are compared: a simple, total N budgeting approach (N balance model) with the inclusion of N losses from denitrification, and the FASSET dynamic model.
SoilKlimaFASSET (50 years) Balance model FASSET rise in leaching compared to the balance model (cummulated over 50 years) FASSET rise in leaching compared to the balance model (cummulated over 200 years) SandyDry7.320.735%64% SandyWet8.420.741%70% Sandy loamDry3.218.318%42% Sandy loamWet5.718.331%58% Comparison of dynamic modelling and budget calculations Additional leaching (kg N ha -1 y -1 ) under different Danish conditions by utilising 120 kg N ha -1 y -1 pig slurry applied every year calculated with respectively FASSET and a balance model. After 50 years, the dynamic modelling predicts a rise in leaching level of 18 to 41 % of the balance model, and even after 200 years the cumulative rise in leaching predicted by the budget model is far from the level predicted by the dynamic model.
The soil pool of organic N is the joker in these calculations! By the end of the 200 year simulation of slurry application, a fairly large amount of organic matter has been lost from the minerally fertilised scenario (lower curve), relative to the scenario with slurry application. Kg N ha -1 The soil N accumulated in the slurry treatment – relative to the mineral fertilised - can of course not be lost as leaching. But even more important, during this period, a fairly large difference in N mineralisation evolves.
Dynamic development The big amounts of N that are exchanged with the soil organic matter (previous slide), and is timing, has bearing influence on the N dynamics: Year -10 -5 0 5 10 15 20 25 30 050100150200 kg N/ha/year Denitrification Soil organic matter Leaching Harvest N entries (sandy loam, wet climate). The curves are cummulated differences between permanent slurry application and mineral fertilisation, calculated with the FASSET model, divided by the number of simulated years. The substitution value of N in slurry for crop utilisation depends on the time frame, and according to the model, the assumed 75% (Danish legislation) are first reached after 100 years. Also note the development in N leaching, due to growing mineralistion differences.
Time frame – lessons learned The initial difference in N leaching calculated by simple N balances and dynamic modelling was quite big, and even after 50 years, only in the order of one third of the expected rise in leaching was simulated. The budget model was carefully tuned, and over an infinite period, the results would have been quite similar. When dealing with changes of agricultural practice that have significant effects on the soil N pool, this example stresses the need to explicitly state a time frame within the political and administrative system. The consequences judged over fx. a 10-year period may be very different than the consequences judged over a very long period. In this example, these differences were up to a factor 4 in estimated leaching.
Biogas utilisation potential in Denmark Danish energy consumption845 PJ Renewable energy117 PJ Biomass (out of the above) 89 PJ (Values for 2005) The scenarios: 1.Utilise 75% of the animal manure for biogas production, yielding 19 PJ extra 2.Combust 75% of the fibre fraction from the residual of biogas, yielding 2.5 PJ extra.
Consequences of biogas for soil organic matter The consequences were modelled over a 50-year period at the national level. Utilising animal manure for biogas production (19 PJ y -1 ) decreases the topsoil C in DK by 3.2%. Combusting the fibre fraction (2.5 PJ y -1 ) decreases the topsoil C in DK by 1.3%. The results are additive, so these two actions will decrease the topsoil C content by 4.5%. The effect will not be equally distributed, so some soils may loose much more organic matter. Note that the decline in C, relative to the energy gained, is markedly larger by the fibre combustion. This is due to the low net energy efficiency of this combustion. So there is a trade-off between the energy gain by fibre combustion and the ease of handling the residues versus the additional loss of soil fertility that needs to be evaluated.
Raw manureBiogas treated manure Biogas treated, adjusted application of mineral fertiliser Biogas treated, fibres combusted, adjusted application Soil carbon changes (kg C ha -1 y -1 ) 78-7 -80 Nitrate leaching (kg N ha -1 y -1 ) 57605452 Soil C and nitrate leaching from a pig farm applied 120 kg manure-N ha -1 Accumulated effects over 50 years So, with unchanged application of N, the N leaching estimated with FASSET has increased slightly when using biogas treated manure. Only when adjusting the supplementary mineral N to obtain the same amount of harvested N as before, the leaching decreases.
Two major reasons: Pollution substitution, as the ammonium (presumably) and denitrification losses decrease when aplying biogass treated manure Reduced build-up of soil N when aplying biogass treated manure – less total C in the digestate. The larger availability of N then causes both plant uptake and leaching to increase – at least judged from the present model parameter settings. Why does the leaching rise, with unchanged levels of nitrogen input?
Conclusions Measurements from the field plot and laboratory can, as demonstrated, by the use of simulation models be projected up to the field, farm, or even regional/national level Simulation models are great tools, and for many issues the only tools at hand. Structure, parameters and results should be examined critically – but please acknowledge the overall usefulness also When looking at changes in pollution from agriculture, the time frame frequently is the thing to consider Biogas production may cause the soil organic matter content in the soil to decrease markedly, especially if the fibre fraction is combusted Treating manure in a biogas plant and spreading the residues with the same amount of N per ha as before probably increases the N leaching slightly, or at least does not decrease leaching. To get noticeable environmental benefit, the amount of supplementary mineral N should be decreased.